基于模型平均方法的基金绩效的研究

发布时间：2018-05-29 09:17

本文选题：基金绩效 + 模型平均方法　；参考：《青岛大学》2017年硕士论文

【摘要】：进入二十一世纪以来,我国的证券投资基金发展态势良好,开放式基金业逐渐发展成基金市场的中流砥柱。基金业绩的评估体系在基金业的发展中起到了举足轻重的作用,如今基金评估方法多样,采用模型选择方法时,选择的依据不够明确。为了避免模型选择产生的偏差,考虑将模型平均方法运用到基金绩效的研究中,探究选取的因素对基金绩效的影响,并期望能改善预测效果。基于模型平均方法的基金绩效的评价研究主要从三个方面进行:首先,搜集对基金业绩产生重大影响的因素,搜集的因素之间存在尽量小的相关性,并且容易构建线性模型;其次,引入S-AIC、S-BIC、JMA和OPT四种模型平均方法,将单只基金的月度净值增长率作为被解释变量,选取三因素模型中的市场因素、规模因素和账面市值比因素,以及另外的持股集中度和基金净值增长率标准差总共五个影响因素作为被解释变量,构建若干个可供选择的线性模型,对每个模型都按照四种不同的标准赋予一定的权重进行实证分析,选取2010年6月之前成立的30只开放式基金作为研究样本,样本评价时间区间为2011年1月1日到2016年3月31日,对2016年4月1日至2016年12月31日共九期的基金月度净值增长率进行预测,将绝对误差和最优率作为预测精度来衡量预测的效果。结果显示,无论从平均绝对误差还是最优率上看,OPT模型平均方法对基金绩效的预测精度最高,预测效果最好,并且具有很好的稳定性,最后基于OPT方法探究出这五个因素都对基金绩效有影响。
[Abstract]:Since the 21 century, China's securities investment funds have developed well, and the open-end fund industry has gradually developed into a mainstay of the fund market. The evaluation system of fund performance plays an important role in the development of fund industry. Nowadays, the evaluation methods of fund are various, and the basis of selection is not clear enough when adopting the model selection method. In order to avoid the deviation caused by model selection, we consider applying the model average method to the study of fund performance, explore the influence of selected factors on fund performance, and expect to improve the forecasting effect. The research of fund performance evaluation based on model average method is mainly carried out from three aspects: firstly, collecting the factors that have great influence on the fund performance, the factors collected are as small as possible correlation, and it is easy to build a linear model; Secondly, four model averaging methods, S-AIC-S-BIC-JMA and OPT, are introduced. The monthly net worth growth rate of a single fund is taken as the explained variable, and the market factors, scale factors and book market value ratio factors in the three-factor model are selected. As well as the other five factors which affect the concentration of shareholding and the standard deviation of net fund growth rate as explained variables, several alternative linear models are constructed. Each model is given a certain weight according to four different criteria, and 30 open-end funds established before June 2010 are selected as the research samples. The sample evaluation time is from January 1, 2011 to March 31, 2016. This paper forecasts the monthly net worth growth rate of the fund for the nine periods from April 1, 2016 to December 31, 2016. The absolute error and the optimal rate are taken as the precision of the forecast to measure the effect of the forecast. The results show that both the average absolute error and the optimal rate of OPT model have the highest forecasting accuracy, the best forecasting effect and the good stability. Finally, based on the OPT method to explore the five factors all have an impact on fund performance.
【学位授予单位】：青岛大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F832.51;F224

【参考文献】